system sluggish, thermal keep frequency at 400MHz
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
thermald (Ubuntu) |
Incomplete
|
Medium
|
koba |
Bug Description
This morning I upgraded to 20.10 from 20.04
The system was quite slow although I have a fast machine. My virtual windows 10 on virtualbox became unusable. When I tried to have the virtual machine open, I could not participate properly in a zoom call (I could still hear the people but they said that my voice was very choppy)
On 20.04 I was super-happy with the speed and I could have as many apps running as I want.
Based on google I started looking at
% journalctl --follow
and this shows quite a few errors but not repeating often enough to explain it.
Then I googled some more and found that /boot/efi was writing and reading.
Then I googled some more and thought I had trouble with gnome. So I reset it to default
% dconf reset -f /org/gnome/
and disabled the extensions. This made things slightly better but by far not acceptable.
After lots of searching I checked the frequency of the CPUs and it was at the minimum 400Hz (as shown by i7z and also other tools). I tried setting the governor with cpufreqctl and similar methods but this did not change anything.
I then found an old bug https:/
% sudo systemctl stop thermald
this seems to work. After a few seconds the frequency shown in i7z goes to ~4500 MHz and the virtual machine seems to work fine.
ProblemType: Bug
DistroRelease: Ubuntu 20.10
Package: thermald 2.3-4
ProcVersionSign
Uname: Linux 5.8.0-25-generic x86_64
NonfreeKernelMo
ApportVersion: 2.20.11-0ubuntu50
Architecture: amd64
CasperMD5CheckR
CurrentDesktop: ubuntu:GNOME
Date: Sat Oct 24 01:39:40 2020
DistributionCha
# This is the distribution channel descriptor for the OEM CDs
# For more information see http://
canonical-
InstallationDate: Installed on 2019-09-27 (392 days ago)
InstallationMedia: Ubuntu 18.04 "Bionic" - Build amd64 LIVE Binary 20180608-09:38
SourcePackage: thermald
UpgradeStatus: Upgraded to groovy on 2020-10-23 (0 days ago)
mtime.conffile.
Changed in thermald (Ubuntu): | |
assignee: | nobody → Colin Ian King (colin-king) |
Changed in thermald (Ubuntu): | |
assignee: | Colin Ian King (colin-king) → Ubuntu Kernel Team (ubuntu-kernel-team) |
Changed in thermald (Ubuntu): | |
assignee: | Ubuntu Kernel Team (ubuntu-kernel-team) → koba (kobako) |
Thermald 2.3-4 has been updated recently to 2.4.3-1ubuntu2 with some relevant upstream fixes. Do you mind verifying if this now addresses the issue. The updates contain the following fixes:
thermald (2.4.3-1ubuntu2) hirsute
* Support Jasper Lake. (LP: #1940629) Jasper- Lake-CPU- model.patch
- 0014-Added-
thermald (2.4.3-1ubuntu1) hirsute
* Pull in bug fixes between 2.4.3 and 2.4.6 (LP: #1931565)
- Disable legacy rapl cdev when rapl-mmio is in use
This will prevent PL1/PL2 power limit from MSR based rapl, which
may not be the correct one.
- Delete all trips from zones before psvt install
Initially zones has all the trips from sysfs, which may have wrong
settings. Instead of deleting only for matched psvt zones, delete
or all zones. In this way only zones which are in PSVT will be
present.
- Check for alternate names for B0D4 device
B0D4 can be named as TCPU or B0D4. So search for both names
if failed to find one.
- Fix error for condition names
The current code caps the max name as the last condition name,
which is "Power_Slider". So any condition more than 56 will be
printing error, with "Power_Slider" as condition name. For example
for condition = 57: Unsupported condition 57 (Power_slider)
- Set a very high RAPL MSR PL1 with --adaptive
After upgrading Dell Latitude 5420, again noticed performance
degradation.
The PPCC power limit for MSR RAPL PL1 is reduced to 15W. Even though
we disable MSR RAPL with --adaptive option, it is not getting
disabled. So MSR RAPL limits still playing role.
To fix that set a very high MSR RAPL PL1 limit so that it never
causes throttling. All throttling with --adaptive option is done
using RAPL-MMIO.
- Special case for default PSVT
When there are no adaptive tables and only one default PSVT table
is present with just one entry with MAX type. Add one additional
entry as done for non default case.
- Increase power limit for disabled RAPL-MMIO
Increase 100W to 200W as some desktop platform already have limit
more than 100W.
- Use Adaptive PPCC limits for RAPL MMIO
Set the correct device name as RAPL-MSR so that RAPL-MMIO can
also set the correct default power limits.
If this fixes the issue please let us know.